giuros01 | be1c017 | 2018-12-04 11:36:08 +0000 | [diff] [blame] | 1 | /* |
Michele Di Giorgio | d9eaf61 | 2020-07-08 11:12:57 +0100 | [diff] [blame] | 2 | * Copyright (c) 2018-2019 Arm Limited. |
giuros01 | be1c017 | 2018-12-04 11:36:08 +0000 | [diff] [blame] | 3 | * |
| 4 | * SPDX-License-Identifier: MIT |
| 5 | * |
| 6 | * Permission is hereby granted, free of charge, to any person obtaining a copy |
| 7 | * of this software and associated documentation files (the "Software"), to |
| 8 | * deal in the Software without restriction, including without limitation the |
| 9 | * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or |
| 10 | * sell copies of the Software, and to permit persons to whom the Software is |
| 11 | * furnished to do so, subject to the following conditions: |
| 12 | * |
| 13 | * The above copyright notice and this permission notice shall be included in all |
| 14 | * copies or substantial portions of the Software. |
| 15 | * |
| 16 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
| 17 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
| 18 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
| 19 | * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
| 20 | * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
| 21 | * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE |
| 22 | * SOFTWARE. |
| 23 | */ |
| 24 | #include "arm_compute/core/NEON/kernels/NETileKernel.h" |
| 25 | |
| 26 | #include "arm_compute/core/Error.h" |
| 27 | #include "arm_compute/core/Helpers.h" |
| 28 | #include "arm_compute/core/ITensor.h" |
| 29 | #include "arm_compute/core/TensorInfo.h" |
| 30 | #include "arm_compute/core/Validate.h" |
| 31 | #include "arm_compute/core/Window.h" |
| 32 | #include "arm_compute/core/utils/misc/ShapeCalculator.h" |
| 33 | |
| 34 | namespace arm_compute |
| 35 | { |
| 36 | namespace |
| 37 | { |
| 38 | Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, const Multiples &multiples) |
| 39 | { |
| 40 | ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output); |
Georgios Pinitas | 3384356 | 2019-12-10 13:33:18 +0000 | [diff] [blame] | 41 | ARM_COMPUTE_RETURN_ERROR_ON(input->data_type() == DataType::UNKNOWN); |
giuros01 | be1c017 | 2018-12-04 11:36:08 +0000 | [diff] [blame] | 42 | ARM_COMPUTE_RETURN_ERROR_ON(multiples.size() > 4); |
| 43 | ARM_COMPUTE_RETURN_ERROR_ON(multiples.empty()); |
| 44 | ARM_COMPUTE_RETURN_ERROR_ON(std::any_of(multiples.begin(), multiples.end(), [](uint32_t e) |
| 45 | { |
| 46 | return e == 0; |
| 47 | })); |
| 48 | |
| 49 | // Validate output if initialized |
| 50 | if(output->total_size() != 0) |
| 51 | { |
| 52 | ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DIMENSIONS(misc::shape_calculator::compute_tiled_shape(input->tensor_shape(), multiples), output->tensor_shape()); |
| 53 | ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output); |
| 54 | } |
| 55 | |
| 56 | return Status{}; |
| 57 | } |
| 58 | } // namespace |
| 59 | |
| 60 | NETileKernel::NETileKernel() |
| 61 | : _input(nullptr), _output(nullptr) |
| 62 | { |
| 63 | } |
| 64 | |
| 65 | void NETileKernel::configure(const ITensor *input, ITensor *output, const Multiples &multiples) |
| 66 | { |
| 67 | ARM_COMPUTE_ERROR_ON_NULLPTR(input, output); |
| 68 | |
| 69 | // Auto initialize output |
| 70 | TensorShape tiled_shape = misc::shape_calculator::compute_tiled_shape(input->info()->tensor_shape(), multiples); |
| 71 | auto_init_if_empty(*output->info(), tiled_shape, 1, input->info()->data_type()); |
| 72 | |
| 73 | // Validate |
| 74 | ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), output->info(), multiples)); |
| 75 | |
| 76 | _input = input; |
| 77 | _output = output; |
| 78 | |
| 79 | // Configure window without padding |
| 80 | Window win = calculate_max_window(*output->info()); |
| 81 | INEKernel::configure(win); |
| 82 | } |
| 83 | |
| 84 | Status NETileKernel::validate(const ITensorInfo *input, const ITensorInfo *output, const Multiples &multiples) |
| 85 | { |
| 86 | ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output, multiples)); |
| 87 | return Status{}; |
| 88 | } |
| 89 | |
| 90 | void NETileKernel::run(const Window &window, const ThreadInfo &info) |
| 91 | { |
| 92 | ARM_COMPUTE_UNUSED(info); |
| 93 | ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this); |
| 94 | ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(INEKernel::window(), window); |
| 95 | |
| 96 | Window output_window{ window }; |
| 97 | output_window.set(Window::DimX, Window::Dimension(output_window.x().start(), output_window.x().end(), _input->info()->dimension(0))); |
| 98 | Window out_slice = output_window.first_slice_window_1D(); |
| 99 | |
| 100 | const auto src_shape = _input->info()->tensor_shape(); |
| 101 | do |
| 102 | { |
| 103 | Iterator output_it(_output, out_slice); |
| 104 | |
| 105 | execute_window_loop(out_slice, [&](const Coordinates & id) |
| 106 | { |
| 107 | const size_t x = id.x(); |
| 108 | const size_t y = id.y(); |
| 109 | const size_t z = id.z(); |
| 110 | const size_t w = id[3]; |
| 111 | Coordinates input_coords{ x % src_shape[0], y % src_shape[1], z % src_shape[2], w % src_shape[3] }; |
| 112 | memcpy(output_it.ptr(), _input->ptr_to_element(input_coords), _input->info()->dimension(0) * _input->info()->element_size()); |
| 113 | }, |
| 114 | output_it); |
| 115 | } |
| 116 | while(output_window.slide_window_slice_1D(out_slice)); |
| 117 | } |
| 118 | } // namespace arm_compute |